AMD's Athlon 64 3400+: Death of the FX-51

by Anand Lal Shimpi on 1/6/2004 4:35 AM EST


Back to Article

  • AtaStrumf - Friday, January 09, 2004 - link

    This is just a quick test not an in-depth review, since I'm sure Anand is rather short on time, but was swept up in all this A64 hysteria and just had to do the test.

    Is's an OK test though, and I really liked the BFBI (bang for buck index). I also like the conclusion, which is one of the more clear ones in recent history. bussines/games/compile --> AMD64, encoding, 3D rendering programs --> Intel. Then just pick your BFBI and you have your CPU.

    Gotta try to keep things simple or else all you get is confusion.

    Maybe reviews should be written like some books, which keep simple information in normal size font, and more detailed in fine print, since it is intended for the more informed and interested readers and would only confuse the rest.
  • Evan Lieb - Friday, January 09, 2004 - link


    Intel's 875P motherboard is at worst 2% slower than ASUS, ABIT, Gigabyte, etc. 875P motherboards. Anand may have used Intel's 875P board merely to guarantee as best he could that no problems would arise during testing, as Intel motherboards are usually the most reliable on the market.

    Take care,

  • EZDEALZ - Thursday, January 08, 2004 - link

    Athlon 64 3200+ processor locked? I have my K8V running at 2.2gHz (220 x 10) with no problems.. Reply
  • TrogdorJW - Wednesday, January 07, 2004 - link

    Still no comment on the use of the Intel motherboard? Oh, well. How about the actual memory timings that the tests ran at? I know the RAM says it's rated for 2-3-2-6 timings, but just because RAM is rated at a certain speed doesn't mean it was actually run at that speed. (Motherboard and chipset issues can often cause problems.)

    As for the suggestion of benchmarking with different memory speeds, johnsonx, I have to think that it would prove to be largely useless. Tests on numerous hardware sites have shown repeatedly that running the memory and system bus asynchronously seriously impairs performance. If you're running a 200 MHz FSB and then OC the memory to 233, you'll probably end up running slower in many applications. Synchronous clocking is the way to go right now, IMO.
  • johnsonx - Wednesday, January 07, 2004 - link

    gee, screwed up again... in my corrected line I left out "controller", as in "memory controller is relatively inefficient"

    But again, I'm guessing you all knew what I meant. I just hate leaving out words...
  • johnsonx - Wednesday, January 07, 2004 - link

    nuts... I didn't finish an edit before I posted. That line in the 6th paragraph should read "this suggests the memory is relatively inefficient in dual-channel mode".

    But I'm guessing you knew what I meant anyway...
  • johnsonx - Wednesday, January 07, 2004 - link

    Seems to me that the extremely close scores between the single and dual channel 2.2Ghz A64 chips suggests:

    1. Any bandwidth benefit of dual-channel memory with the A64FX is offset by the latency penalty of registered memory. With the memory controller on-die, latency is apparently everything.

    2. The single-channel version of the A64 memory controller is extremely efficient.

    3. The dual-channel version of the A64FX memory controller is perhaps somewhat inefficient, or perhaps the cpu-core to memory controller interconnect is not as fast as it should be. Latency penalty aside, all that extra bandwidth should amount to a bigger performance gain.

    Our intrepid friends at Anandtech could shed some light on this by testing the 2.2Ghz chips with different memory speeds. I believe both chips will support memory speeds down to 133Mhz/PC-2100.

    This might provide interesting results. If the numbers increase fairly linearly with memory speed, but are relatively close between single and dual channel, this suggests the memory controller is fairly relatively in dual-channel mode. If the numbers for single channel scale much more with memory speed than with dual-channel (i.e. if single DDR-266 is much slower than single DDR-400, but dual DDR-266 (with roughly DDR-533 equivalent bandwidth) is not much slower than dual DDR-400), then that suggests that the A64 is architecurally limited somehow, and at current speeds is well fed by 3200Mb/s bandwidth, (or whatever the tests show), dual-channel or not.

    A related interesting test would be to repeat the same set of tests using an A64 and A64FX running at the same 2.2Ghz, but using a 220Mhz clock speed and a 10x multiplier. For this, obviously an A64 3200+ would be overclocked... This would tell us whether the memory controller's performance (and the performance of the on-die link between the core and memory controller) is dependant on the base clock speed.

    just my lofty ideas I can't test myself since I don't have any A64 at all....
  • Icewind - Wednesday, January 07, 2004 - link

    Fudge.....I really wanted Dual Channel with my Athlon FX next summer, but god I just can't swing a $750 cost, I wanna get a R420 also. Damnitm why cant AMD give the A64 dual and just let the FX have the unlocked multiplier for ethusisasts?

    For some reason, Tejas is starting to sound pretty good now
  • Wesley Fink - Wednesday, January 07, 2004 - link

    In my benchmarking of the FX versus the 3400+, the FX is the clear winner in Workstation Performance, Media Encoding, and some of the more demanding games. The point is not whether the FX is faster, because it is in some applications, it is whether it is now worth the 40% higher price asked for it.

    The P4 wants all the bandwidth you can give it, and there is a big difference in perfromance between Dual-Channel and Single-Channel memory. The A64 architecture is not as dependent on Memory Bandwidth for performance. Having said that, Dual-Channel on the A64 is faster in some applications and it may be even more important to 64-bit performance.

    The upcoming Socket 939 IS dual-Channel, so we will be seeing more Dual-Channel solutions from AMD.
  • DAVIDS - Wednesday, January 07, 2004 - link

    Insomniac: yes, what you say is true. But is no consistency to the way Anandtech tests CPU's. Look at the benchmarks used for the 3000+ and the original Athlon64 review from 9/23. Also, the FX does not have a multiplier lock, a fact this article seems to downplay. Anyone who overclocks understands the importance of an unlocked CPU. Reply
  • Insomniac - Wednesday, January 07, 2004 - link

    DAVIDS: Those are real important numbers. I know my favorite applicatiomn is Sandra and game is 3DMark...

    Seriously, those are just theoretical. It's nice to know what a processor CAN do, but it is much better to see what it ACTUALLY does.
  • Insomniac - Wednesday, January 07, 2004 - link

    Thanks Jason! Reply
  • DAVIDS - Wednesday, January 07, 2004 - link

    It seems like Anandtech chose to ignore certain benchmark tests like Sandra or 3Dmark where the FX will clearly outclass the 3400+. If you look at the benches done by Amdzone, the FX annihilates the 3400+ in any test involving memory bandwidth. Reply
  • mkruer - Wednesday, January 07, 2004 - link

    Last time I checked the "NewCastle" core was suppose to have a new memory controller, that would allow for dual channel. This core is not the NewCastle it’s just AMD pulling some of their under spec’ed chips disabling 1/2 the cache and re-branding it as a different model. Usually I would have a huge issue with this, but after looking at a bunch of benchmarks, I see the 1/2 the cache is not a big hindrance at all. All the information that I have and seen says the NewCastle core will be the 130nm equivalent of the Paris core, that’s it.

    But on top of that what I am surprised about was no mention that the AMD64 is handy caped at the moment. I am sure once 64-bit OS comes out and 64-bit Apps are available all the “multimedia” wins that the Intel chips currently have will instantly flip. Rule of thumb says that “multimedia” will benefit the most from the increase in bits, and in the case of 3DSMAX don’t be too surprised if you see 100%+increase in performance. Don’t believe me take a look at the LAME encoding on 64-bit OS with 64-bit lame.
  • mkruer - Wednesday, January 07, 2004 - link

  • destaccado - Wednesday, January 07, 2004 - link

    well at least he didn't publish anything until page 2 this time :) Reply
  • raskren - Wednesday, January 07, 2004 - link

    Please don't feed the trolls. Reply
  • CRAMITPAL - Tuesday, January 06, 2004 - link

    I must ask the obvious because it clearly escaped most folks here. Do you think AMD would release the 3400+ and kill FX51 sales and profits if it didn't have an FX53 ready to roll next month when Pisscott the Flame Thrower paper launches??? You folks must think this is AMD's first day on the job or something. You may recall AMD is the same company who made the PIII obsolete (4) years ago with the launch of Athlon and AMD made the Piss 4 and Xeon obsolete (6) months ago with Opteron/A64...

    If only you knew! :>))
  • TrogdorJW - Tuesday, January 06, 2004 - link

    Hey, is it just me, or is the Intel system rather handicapped by running on an Intel D875PBZ motherboard? From what I've seen on benchmarks, that board routinely finishes 2 to 10% slower than the top enthusiast boards. Given that Anandtech in the past has praised the Asus P4C800 (and P4C800-E), not to mention the A-Bit IS7 and IC7-MAX, why use the standard Intel 875 motherboard?

    No, it wouldn't help the P4 win the overall title. But it would have been a more accurate portrayal of the top performance we can expect from the systems. A 5-10% increase on several of the benchmarks would have closed the gap between A64 and P4.

    The end result, of course, is that the 3400+ still beats the best P4s in games and scientific work. With office type applications, it's probably pointless to argue for one or the other, since they're so fast. And the P4 still wins on 3D rendering and media encoding.

    So it's not a really major flaw, but I still have to ask: Why? Did you really have no other P4 enthusiast motherboards available? And yet you had an Intel motherboard. Seems a little odd, at best. It's the type of thing that makes people wonder.
  • edub82 - Tuesday, January 06, 2004 - link

    The message is clear!!! AMD and Intel both offer great CPUs at similiar price points. Reply
  • Pumpkinierre - Tuesday, January 06, 2004 - link

    #15 I have'nt heard AMD call the 3000+ Newcastle and other sites dont refer to it as such. Many sites say that the die size and transistor count is the same (193mm2, 105million) as the 3200+. hardtecs4u has cpz on the 3000 and 3400:

    Same family, same stepping, same revision and code name:clawhammer. I havent found out whether cache associativity has been cut down from 16 but that is a minor point.

    If it looks the same and smells the same dont be prudish call it the same.
  • JohnrrDrake - Tuesday, January 06, 2004 - link

    I appreciate the VisualStudio compiler test.

    As a developer, I am much more interested how much compile time is saved instead of how many more FPS I get.

    You may also consider throwing in a GCC compiler test (linux kernal is fairly typical).

  • KF - Tuesday, January 06, 2004 - link

    >reljam is right, if it isn't CPU limited,
    >why include it? Or why not lower the res?
    Because it tells the truth about what people should expect? Gamers might like to know that they can use a slower (and cheaper) CPU. Very few people even have a graphics card as good as the testers used, so they don't even need CPUs like these. The real problem with benchmarks is how (or whether) they apply to real use. That makes null results like this important, IMO. Boring maybe. But useful.

    Older games were included, and they show a CPU difference. I suppose that is because the newer games use the GPU for things that older games did with the CPU.

    >But I think AMD may have shot themselves in the foot...

    Companies that are not in a monopoly position need to put forth the best product they can at the time, or else get crushed by the competition. If that means some products are short-lived...well it's better than losing. A few months at the top is actually not bad the way things go. If you won't beat yourself, then the competition will.

    I remember Intel putting out pin incompatible PPGA, FCPGA, and FCPGA2 PII/IIIs in quick succession, and then the totally incompatible P4. That was Intel scrambling with AMDs close competition.
  • KristopherKubicki - Tuesday, January 06, 2004 - link

    By AMD's definition, NewCastle is the same core as ClawHammer with half the cache. Regardless if it is a compeltely different core or not the performance is going to be the same between a 1/2 ClawHammer or a NewCastle.

  • dvinnen - Tuesday, January 06, 2004 - link

    reljam is right, if it isn't CPU limited, why include it? Or why not lower the res?

    And are yall sure this is newcastle? Always figured it was a clawhammer with half the cache turned off. If you take the heat spreader off, it should be easy to tell. I seeing the bang-for-buck comparisons in a few weeks after the price drops start to take effect would be nice.

    atir: they did some 64 bit to 32 comparisons in one of the opteron reviews.
  • atlr - Tuesday, January 06, 2004 - link

    Has anyone seen any performance comparisons of 32-bit versus 64-bit compiled programs? Reply
  • reljam - Tuesday, January 06, 2004 - link

    The AquaMark DX9, Halo (both benchmarks), and GunMetal are graphics limited benchmarks and are not adding any value to the review. Reply
  • Lonyo - Tuesday, January 06, 2004 - link

    Why run the games at 1024x768? I know it gives more real world performance, but if a game is limited by the graphics card, what use does the benchmark have in a CPU article (like the Aquamark benchmark numbers (not the CPU part though)
    Things such as UT2k3 are valid, and Comanche 4 woul dbe very good to look at differences (since unless I am mistaken, it used to be CPU limited most of the time).
    Of course, time constraints may be the issue, but it seems in a way wasteful to do tests which are more GPU tests than CPU tests.
    But otherwise, very good article in terms of non-gaming stuff, and it shows that the 3000+ is probably the best buy of the lot, good performance and a nice price.
  • EddNog - Tuesday, January 06, 2004 - link

    Well on NewEgg when U go to buy your CPU, check the memory (Single channel DDR, vs. Dual channel DDR), the number of pins/socket type (Socket 754, Socket 939) and most importantly, the cache size (512KB, vs. 1MB).

  • Icewind - Tuesday, January 06, 2004 - link

    I wish they would give the freaking newcastle Athlons a different name, cause how the hell are you supposed to tell the difference cause they have the samen damn name??

    Sometimething tells me the Athlon FX 939 pin next year will change things.
  • Jeff7181 - Tuesday, January 06, 2004 - link

    I found fault with the article... no FS2004 benchmark. Can I have it please? :D Reply
  • Jeff7181 - Tuesday, January 06, 2004 - link

    I agree with everyone so far =)

    But I think AMD may have shot themselves in the foot by releasing the 3400+, which performs exactly the same as the insanely priced FX-51. Unless they have some tricks up their sleeve with socket 939 that will improve performance, why would anyone spend twice the money on the FX-51 when the 3400+ provides 98% of the performance of the FX-51?
  • Jason Clark - Tuesday, January 06, 2004 - link

    The print article issue is fixed.

  • AlexWade - Tuesday, January 06, 2004 - link

    I liked the compile times benchmark! Please have it all new reviews of CPU's. Reply
  • PrinceXizor - Tuesday, January 06, 2004 - link

    Nice review! I especially liked the price/performance charts. It should be interesting to see how AMD handles the transition from 754 to 939 sockets.

    The thing I find most impressive is that AMD is staying ahead of Intel as far as performance, something many of us did not think possible given Hammer's seemingly endless wait.

    Intel had to rush out an EE version to remain competitive while it waited for Prescott (I'm not intel bashing, I'm sure Prescott will compete nicely).

    In the meantime, AMD is the one hitting the "3400+" performance arena before the 3.4Ghz Prescotts hit. The question always was, can AMD execute with the Athlon 64 as well as they did with the Athlon XP? The answer seems to be yes. Very well done AMD!

  • Insomniac - Tuesday, January 06, 2004 - link

    I meant hints. ;) Reply
  • Insomniac - Tuesday, January 06, 2004 - link

    Nice article. Any hits on how the battle will look when Prescott hits? :)

    I noticed in the print view, none of the charts are showing up.
  • FearoftheNight - Tuesday, January 06, 2004 - link

    great review...hope to see socket 939 coverage coming soon :D Reply

Log in

Don't have an account? Sign up now